LLM inference in C/C++
C++ ggml
spec: fix failed to decode mtmd chunk with DFlash (#28587)
Jesus Gulfofa6769818708
diff --git a/common/speculative.cpp b/common/speculative.cppindex 2db381d580..b7811b853e 100644--- a/common/speculative.cpp+++ b/common/speculative.cpp@@ -1094,8 +1094,7 @@ struct common_speculative_impl_draft_dflash : public common_speculative_impl {// Target prefill may contain token IDs or multimodal embeddings. Both// produce the target-layer features used to seed the draft KV cache, so- // skipping the embedding batches leaves a hole in the draft's cache and- // the next injection fails to initialize.+ // embeddings are injected too, except the pinned ones skipped below.// TODO: revisit after https://github.com/ggml-org/llama.cpp/pull/24669 is mergedconst bool has_tokens = batch_in.token != nullptr;const bool has_embeddings = batch_in.embd != nullptr;@@ -1131,6 +1130,13 @@ struct common_speculative_impl_draft_dflash : public common_speculative_impl {}const int32_t n_rows = i_batch_end[seq_id] - i_batch_beg[seq_id] + 1;+ // an M-RoPE image pins all its rows to one position, so a windowed draft+ // cache cannot free cells for it - skip it, the draft can jump over the gap+ const bool pos_pinned = batch_in.pos[i_batch_beg[seq_id]] == batch_in.pos[i_batch_end[seq_id]];+ if (has_embeddings && n_rows > 1 && pos_pinned) {+ continue;+ }+for (int32_t offset = 0; offset < n_rows; offset += n_ubatch) {const int32_t n_chunk = std::min(n_ubatch, n_rows - offset);